Re: [PATCH] oprofile/loongson2: rename cpu_type from godson2 to loongson2

[Robert Richter <robert.richter@am...>]

On 21.10.09 22:51:46, Wu Zhangjin wrote:
> This patch try to unify the naming method between kernel and the
> user-space oprofile tool. 'Cause loongson is used instead of godson in
> most of the places, and just confer with the developer of the user-space
> tool, we are agreed to use loongson instead, which will help a lot to
> the future maintaining.
> 
> (This patch is very important to help the user-space support upstream,
>  so, Ralf, could you please merge it into your -queue branch, thanks!)
> 
> Signed-off-by: Wu Zhangjin <wuzhangjin@...>

Acked-by: Robert Richter <robert.richter@...>

Ralf, please apply to your tree.

Thanks.

-Robert

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@...

[PATCH] oprofile/loongson2: rename cpu_type from godson2 to loongson2

[Wu Zhangjin <wuzhangjin@gm...>]

This patch try to unify the naming method between kernel and the
user-space oprofile tool. 'Cause loongson is used instead of godson in
most of the places, and just confer with the developer of the user-space
tool, we are agreed to use loongson instead, which will help a lot to
the future maintaining.

(This patch is very important to help the user-space support upstream,
 so, Ralf, could you please merge it into your -queue branch, thanks!)

Signed-off-by: Wu Zhangjin <wuzhangjin@...>
---
 arch/mips/oprofile/op_model_loongson2.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/mips/oprofile/op_model_loongson2.c b/arch/mips/oprofile/op_model_loongson2.c
index 9536b21..08d4b09 100644
--- a/arch/mips/oprofile/op_model_loongson2.c
+++ b/arch/mips/oprofile/op_model_loongson2.c
@@ -22,7 +22,7 @@
  * otherwise, the oprofile tool will not recognize this and complain about
  * "cpu_type 'unset' is not valid".
  */
-#define LOONGSON2_CPU_TYPE	"mips/godson2"
+#define LOONGSON2_CPU_TYPE	"mips/loongson2"
 
 #define LOONGSON2_COUNTER1_EVENT(event)	((event & 0x0f) << 5)
 #define LOONGSON2_COUNTER2_EVENT(event)	((event & 0x0f) << 9)
-- 
1.6.2.1

Re: [PATCH-V2] adds support for loongson2

[John Levon <levon@mo...>]

On Wed, Oct 21, 2009 at 10:23:14PM +0800, Chen Jie wrote:

> I've ran 'make distcheck', no error reports.
> Currently only a few parts in kernel source reference "loongson" as
> "godson", these will be changed to "loongson" in the future.

Then you need to finish fixing the kernel before trying to work around
it in userspace, unless you can convince us otherwise...

regards
john

Re: [PATCH-V2] adds support for loongson2

[Chen Jie <chenj@le...>]

2009/10/21 Maynard Johnson <maynardj@...>:
> Chen Jie wrote:
>> The second version of patch adds oprofile userland support for ICT
>> loongson2 processor.
>> Signed-off-by Chen Jie <chenj@...>
>>
>> == changes since patch v1 ==
>> * Fix 'make distcheck' fails
>> * Add new entry to ChangeLog
>>
>> About godson2 and loongson2:
>> godson is the codename, and loongson is the trademark, using
>> "loongson" is perferred.
>> The relationship may like Nehalem and core i7.
>>
>> Currently, the kernel identifies loongson2 as "mips/godson2"
>> (arch/mips/oprofile/op_model_loongson2.c), so I put the substantial
>> content in godson2 events, and let loongson2 events include it.
> This seems wrong.  The way you explain it, this is *not* analogous to Nehalem.  For the Nehalem-based processors, the oprofile kernel driver reports their cpu_type as the specific type; e.g. "i386/core_i7".  The event info is put into the events/i386/nehalem/events file, and then events/i386/core_i7/events file simply includes the nehalem events file.  If the kernel reports the processor as "mips/godson2", then all you need is the godson2 events file.  I don't see where the loongson2 events file would ever come into play.  Did you run 'make distcheck'?  This must pass before your patch can be accepted.

I've ran 'make distcheck', no error reports.
Currently only a few parts in kernel source reference "loongson" as
"godson", these will be changed to "loongson" in the future.
>
> Regards,
> -Maynard
>
>>
>> Thanks,
>> -Chen
>>
>
>

Re: [PATCH-V2] adds support for loongson2

[Maynard Johnson <maynardj@us...>]

Chen Jie wrote:
> The second version of patch adds oprofile userland support for ICT
> loongson2 processor.
> Signed-off-by Chen Jie <chenj@...>
> 
> == changes since patch v1 ==
> * Fix 'make distcheck' fails
> * Add new entry to ChangeLog
> 
> About godson2 and loongson2:
> godson is the codename, and loongson is the trademark, using
> "loongson" is perferred.
> The relationship may like Nehalem and core i7.
> 
> Currently, the kernel identifies loongson2 as "mips/godson2"
> (arch/mips/oprofile/op_model_loongson2.c), so I put the substantial
> content in godson2 events, and let loongson2 events include it.
This seems wrong.  The way you explain it, this is *not* analogous to Nehalem.  For the Nehalem-based processors, the oprofile kernel driver reports their cpu_type as the specific type; e.g. "i386/core_i7".  The event info is put into the events/i386/nehalem/events file, and then events/i386/core_i7/events file simply includes the nehalem events file.  If the kernel reports the processor as "mips/godson2", then all you need is the godson2 events file.  I don't see where the loongson2 events file would ever come into play.  Did you run 'make distcheck'?  This must pass before your patch can be accepted.

Regards,
-Maynard

> 
> Thanks,
> -Chen
> 

[PATCH-V2] adds support for loongson2

[Chen Jie <chenj@le...>]

The second version of patch adds oprofile userland support for ICT
loongson2 processor.
Signed-off-by Chen Jie <chenj@...>

== changes since patch v1 ==
* Fix 'make distcheck' fails
* Add new entry to ChangeLog

About godson2 and loongson2:
godson is the codename, and loongson is the trademark, using
"loongson" is perferred.
The relationship may like Nehalem and core i7.

Currently, the kernel identifies loongson2 as "mips/godson2"
(arch/mips/oprofile/op_model_loongson2.c), so I put the substantial
content in godson2 events, and let loongson2 events include it.

Thanks,
-Chen

Re: Install OProfile-0.95 with popt-1.13 & binutils-2.19.51.0.14 but failed

[nick 314 <nick314@gm...>]

Dear Maynard:

Thanks for your response. Sorry for I didn't describe my installation
environment in details.
I did install to some other location but not /usr/local, so I need to
configure OProfile with --with-extra-libs & --with-extra-includes. But now,
I solve this problem.
I forgot to verify --with-extra-includes argument before, which was
/home/nick314/fakeroot/lib/, but it should be
/home/nick314/fakeroot/include/.

Sorry for my mistakes, and appreciate your response again. Thanks a lot.

Below is my installaton commands.

popt
./configure --with-kernel-support --host=mips-linux-gnu
--prefix=/home/nick314/fakeroot/
make
make install

binutils devel
./configure --with-kernel-support --host=mips-linux-gnu
--prefix=/home/nick314/fakeroot/ --enable-shared
make
make install

oprofile
./configure --with-kernel-support --host=mips-linux-gnu
--prefix=/home/nick314/fakeroot/
--with-extra-libs=/home/nick314/fakeroot/lib/
--with-extra-includes=/home/nick314/fakeroot/include/
--with-binutils=/home/nick314/fakeroot/
make
make install

nick314

On Wed, Oct 21, 2009 at 2:02 AM, Maynard Johnson <maynardj@...>wrote:

> nick 314 wrote:
>
>> Dear all:
>>
>> I tried to install OProfile-0.95 by mips toolchain.
>> But it said that I need to cross compile popt & binutils(bfd, iberty) in
>> order to install OProfile.
>> Then I download source codes of popt and binutils, and compile them.
>> After completion, I didn't find out libiberty.so.
>>
> I presume you did a standard 'configure' and 'make install', resulting in
> an installation of these packages to /usr/local.  Is that correct or did you
> install to some other location?  I'm no expert in cross-compile situations,
> so maybe someone else on the list can help out.  But assuming you did *NOT*
> install binutils to /usr, then you need to use the '--with-binutils'
> configure option to tell the oprofile build to use the binutils in
> /usr/local (or wherever you installed it).  But there is no configure option
> for an alternate install location for popt.  Since you apparently didn't get
> a configure error for popt, then I wonder what libpopt was found --
>  AC_CHECK_LIB, by default, will only check /usr.
>
> -Maynard
>
>>
>> So there was a problem when I compile OProfile with error message like
>>
>> /opt/mips-4.3/bin/../lib/gcc/mips-linux-gnu/4.3.3/../../../../mips-linux-gnu/bin/ld:
>> opimport.o: Relocations in generic ELF (EM: 3).
>>
>> I do configure binutils with --enable-shared, but in vain.
>> By the way, I download binutils from
>> ftp.kernel.org/pub/linux/devel/binutils/. But I don't know whether it is
>> binutils-devel or not?
>>
>> I have no idea about how to solve this problem.
>> Would someone could give me directions or solutions? Thanks a lot.
>>
>> Best regards,
>> nick314
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>>
>> ------------------------------------------------------------------------------
>> Come build with us! The BlackBerry(R) Developer Conference in SF, CA
>> is the only developer event you need to attend this year. Jumpstart your
>> developing skills, take BlackBerry mobile applications to market and stay
>> ahead of the curve. Join us from November 9 - 12, 2009. Register now!
>> http://p.sf.net/sfu/devconference
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> oprofile-list mailing list
>> oprofile-list@...
>> https://lists.sourceforge.net/lists/listinfo/oprofile-list
>>
>
>


-- 
I am Nick314

Fwd: Help for Xeon Quad core 64 bit, Xen kernel

[Nishikant Deshmukh <nishikant.deshmukh@gm...>]

---------- Forwarded message ----------
From: Nishikant Deshmukh <nishikant.deshmukh@...>
Date: Tue, Oct 20, 2009 at 11:23 AM
Subject: Help for Xeon Quad core 64 bit, Xen kernel
To: oprofile-list@...


Hi folks,

  I am using Xeon Quad core 64 bit on a Xen kernel.
When I run ophelp -r I get *CPU with timer interrupt*
Does it mean my processor is not supported? or do I need to set some other
parameters?

I am using oprofile 0.9.3

Kernel version Linux nishikant 2.6.26-2-xen-amd64 #1 SMP Thu Aug 20 00:36:34
UTC 2009 x86_64 GNU/Linux

-- 
Best Regards,
Nishikant Deshmukh
Graduate Student
Johns Hopkins University Homewood Campus

Sent from Baltimore, MD, United States



-- 
Best Regards,
Nishikant Deshmukh
Graduate Student
Johns Hopkins University Homewood Campus

Sent from Baltimore, MD, United States